Ep. 41 - Awakening from the Meaning Crisis - What is Rationality?



my notes

  • Stannovich: rationality reliable and systematic overcoming of self-deception, and the potential affording of flourishing by optimizing process of achieving our goals
  • Caveat: as try to optimize often change the goals we are pursuing, appreciate the value of the process not just the end result of the process
  • Rationality debate: experiments seem to reliably show:
    • That people acquiesce acknowledge authority of how should reason
    • Reliably fail to meet those standards
  • One interpretation is that most people are irrational in nature. But this has important implications for moral, legal, political, developmental status -> rationality is deeply existential
  • Cohen: human beings aren’t comprehensively irrational, because standards of rationality come from them

(theist would say the standards come from God)

Cohen and Stannovich

  • Cohen:
    • At the level of my competence, my competence contains all the standards.
    • Can reflect, and get at the underlying competence.
    • What I’m doing when propose normative theory, giving an account of the competence they possess and demand they do their best to reduce the performance errors and meet that confidence
    • At the level of our competence we are fundamentally rational. According to Cohen
    • Cohen says dismiss these experimental results because we have the competence
  • Stannovich and West reply:
    • If Cohen right, errors are all performance errors: competence performance distinction
    • Ex: child, competence not sufficiently developed, why do we think it’s about competence -> errors that reflect defect in competence are systematic errors (Piaget), across different contexts
    • Performance errors not systematic - circumstantially driven
    • How do we see if errors are systematic: If make error X, highly predictive of other errors
    • We can see if make a failure in critical detachment does that mean I’ll also tend to show belief for severence, or tend to leap to the wrong conclusion?
    • Answer is yes: the errors we make are systematic
    • Errors at the level of competence.

Multiple Competencies

  • But there’s something right about Cohen’s argument. Stannovich acknowledges it. We have to be the source of our standards, but the conclusion that performance errors is wrong. How put this together?
  • Step back and look at assumption of Cohen’s argument. Cohen assuming:
    • that it’s a single competence
    • That it’s static
    • That it’s individualistic
  • Platonic idea: what if have two competencies both working to achieve my goal, but may conflict with each other
  • That would mean I’m the source of all the standards, but at the level of competence can be in error because in conflict with each other
  • We don’t have a single competency
  • That’s why should not assume and uncritically assume that can reduce rationality and identify it with the single competence of syllogistic reasoning - that’s wrong, doesn’t follow the science
  • Cohen assuming competence full blown, static, done -> not always fully developed.

(in other words: rationality is a skill to be developed and practiced)

  • This way Stannovich can say Cohen’s argument fundamentally right, though wrong on the hidden premise that the competence is single and static, we have multiple competencies that are ongoing and developing


  • Different approach: ought implies can
  • Are we applying the right normative theory to these experiments when judging them to be irrational
  • We cannot consider all of our assumptions, combinatorially explosive.
  • We can’t be algorithmic, can’t work in terms of certainty/complete -> if tried to be comprehensively deductive combinatorially explosive, cognitive suicide, undermines any attempt to achieve goals
  • But don’t just arbitrarily choose -> pick the relevant implications, relevant contradictions, etc.
  • Stannovich agrees with that
  • In the experiments using formal logic, formal probability theory -> scientist in the experiment using formal theories that can only be applied in very limited contexts.If tried to apply them comprehensively in my life, doomed to fail - ought that can’t possibly meet
  • Wrong normative theory
  • Doesn’t mean can ignore logic, it’s about when where and to what degree should use logic in a formal manner
  • Stannovich and West reply:
    • Say it’s right but not about rationality. Computational limitations is actually not about rationality, it’s about intelligence.
    • Show us that there is a deep difference between being intelligent and rational
    • Foolishness can be from being highly intelligent and irrational
    • Test of computational limitations is test of intelligence
    • Intelligence: capacity to deal with computational limitations/capacity to do relevance realization
    • Have a way of measuring capacity to do computational limitations (g)
    • All of the reasoning tasks also form a strong positive manifold (gr): general factor of reasoning.
    • If Cherniak is right rationality and intelligence would be equal and would be a relationship between how intelligent you are and how well do on these experiments -> doesn’t happen, at best 0.3
    • Intelligence is necessary but not sufficient for being rational
    • Just being intelligent, just being able to use logic not enough for rationality
    • Very processes that make one intelligent can make one irrational
  • What’s the missing piece for rationality?


  • Pointed out something that Stannovich takes seriously
  • Difficulty of interpreting the experiments
  • Rely on distinction between a fallacy and a misunderstanding
  • Two ways can give the wrong answer:
    • Fallacy: Interpret the problem correctly, but reason incorrectly -> wrong answer
    • Misunderstanding: reason correctly, but understood the problem incorrectly
    • One have the incorrect interpretation nothing wrong with reasoning
  • If want to conclude people are irrational have to conclude fallacious cognition, not some kind of distortion in the communication
  • But this is difficult - they aren’t independent enough to cleanly interpret the results
  • Preliminary account of understanding:
    • To understand X, we ask them to give us something identical to X
    • Ask to give something that contradicts X
    • Give something that X implies
    • Something that is relevant to X
    • What is X relevant to
  • Problem: puts aside relevant to X, looks at first three. Someone understands us if they reason the way that we do. Scientists assuming the participants have understood the problem correctly, and reasoned correctly.
  • If they’ve understood the problem correctly then they reason the way the scientist does, but they reason in way scientists doesn’t when try and solve problem
  • Fact that consistently getting wrong answer may be evidence they are misunderstanding the question
  • Or maybe miscommunicated the problem -more problematic
  • Stannovich:
    • need a normativity construal: normative on how people interpret, make sense of size up, how they formulate the problem - has to be independent of inferential norms
    • Means there’s a noninferential aspect to rationality, deals with understanding - relevance
    • Relevance is pre-inferential
    • We have standards on what good problem formulation is, vs. bad -> insight problem-solving. Bad = puts you into combinatory explosive search space, not paying attention to how salience is misleading you.
    • Insight is crucial to being rational.
  • Propose that we need to understand both the role of insight and inference in rationality - much more problematic
  • Rationality and wisdom starting to overlap more and more
  • Give up on rationality being like Mr/. Spock, or just being really smart, start getting into the problematic notion of rationality, get into multiple-competencies, normativity on construal, the generation of insight, not on the generation of inference
  • Constual doesn’t play a significacnt role Stannovich
  • Missing piece for Stannovich:
    • Intelligence not equal rationality: 0.3
    • What accounts for most of the variance
    • Argues for a cognitive style, bad mindware, psychotechnology
    • Psychotech using: can have poor mind-ware (like software)
    • Appropriate cognitive style: something you can learn, learn a set of sensitivities and skills
    • Cognitive style most predictive of doing well on reasoning test is: active open-mindedness -> to train yourself to look for these patterns of self-deception, look for biases

(Oooh! This is my jam!)

  • Biases:
    • Confirmation bias - look for info that confirms beliefs
    • Essentialism bias - treat any category as pointing to an essence shared by all the members
    • Misusing the availability heuristic - judge somethings probability by how easy can remember it
    • Bias = misusing a heuristic
  • Can learn about these and sensitise myself to looking for these biases in my day-to-day cognition, then I actively counteract them.

(OOOH! Again, this is my whole jam!)

  • I can ask you to help me overcome my confirmation bias

(Ok, is he copying from me now! I say this all the time!)

  • Practice with you, can start to internalize you and get better at finding my instances
  • Baron: but don’t overdo this, will choke on the combinatory explosion. Do to the right degree. Becomes more nebulous - wisdom
  • Need for cognition predicts active open-mindedness

(Ding ding)

  • Add: curiosity, want to solve problems, not just gather facts.
  • Also: wonder. How much it opens you up. Entire worldview
  • Rationality existential issue

(this episode was quite validating!)